Since last year, interest in generative artificial intelligence (GAI) has surged. This advanced technology that can create new, human-like content without specific programming. Although GAI has existed for a while, progress in training data volume, training techniques, and computational power have led to GAI outputs that closely mirror content created by humans. Take OpenAI’s ChatGPT, for instance. This conversational AI exemplifies the capabilities of GAI by producing a range of content, from essays to programming code, all in response to user prompts. While technologies like ChatGPT were not originally designed for educational purposes, increasing evidence indicates that students (and instructors) are using them for a variety of teaching and learning purposes. Nonetheless, there is a limited understanding and empirical evidence regarding how these technologies support student learning. Furthermore, despite ongoing concerns about potential misuse, such as cheating or uncritical adoption, both anecdotal and empirical evidence suggest that:
1) students appreciate and regularly use generative AI technologies like chatbots powered by large language models (LLMs) for their learning;
2) educators recognize that these tools are here to stay and resisting them is futile;
3) educators are seeking solutions on how to best guide students in using these tools to support rather than hinder their learning; and
4) students are also concerned about how to optimize the benefits of these technologies while progressing towards a deeper understanding and developing essential competencies.
Building on this context, the Centre for Excellence in Biology Education (bioCEED) and the Centre for the Science of Learning & Technology (SLATE) at the University of Bergen have taken a forward-thinking approach. Together, they have been researching the integration of large language models at the University of Bergen. The project, GAILE (Generative AI in Learning) has three main goals:
1) facilitate the exchange of experiences with large language models (LLMs) like ChatGPT among instructors;
2) investigate the benefits and challenges these models offer for teaching; and
3) collaboratively brainstorm creative ways to include LLMs in courses to enhance student learning.
On August 22, 2023, bioCEED, in partnership with SLATE, organized a day-long workshop that gathered 19 educators from various departments within the faculty of Mathematics and Natural Sciences.
In group sessions, attendees brainstormed the advantages and potential hurdles of using LLMs. The group noted that current tools like ChatGPT, Bard, and Bing can summarize texts, draft outlines, serve as personal tutors, provide feedback, suggest code, and brainstorm. The group also recognized challenges with LLMs, such as crafting the right prompts, ensuring all students have equal access, managing biases, dealing with concerns about cheating, and verifying the accuracy of AI outputs.
During the workshop, instructors also worked together to draft plans for incorporating an LLM into their upcoming courses. These plans were gathered for future reference, and ongoing weekly check-ins with participants are set to monitor progress. A follow-up meeting is scheduled for early December to share experiences and outline the next steps for this initiative. More updates will be shared soon!
Project leaders conducted an early study of instructor and student perceptions, across the Faculty of Mathematics and Natural Sciences, in late February 2023. Findings from this initial survey led to an essay in Science Norway and the workshop discussed above. A second survey, of biology-related study programs across Norway, was conducted in October 2023. This larger survey was designed to meet the needs of many bioCEED projects, and a few LLM-related questions were included. Findings from this second survey will be shared soon.
Project Period:
February 2023 -
Project Period:
Funded By:
bioCEED & SLATE
Project Leader:
Department of Biological Sciences, University of Bergen: Anne E. Bjune, Ståle Ellingsen, and Sehoya Cotner & Department of Informatics, University of Bergen: David Grellscheid
Project Members:
SLATE, University of Bergen: Raquel Coelho, Barbara Wasson
Project Partners:
